CN114175071A - Intelligent data pre-processing techniques for facilitating load shape prediction for utility systems - Google Patents

Intelligent data pre-processing techniques for facilitating load shape prediction for utility systems Download PDF

Info

Publication number
CN114175071A
CN114175071A CN202080041162.3A CN202080041162A CN114175071A CN 114175071 A CN114175071 A CN 114175071A CN 202080041162 A CN202080041162 A CN 202080041162A CN 114175071 A CN114175071 A CN 114175071A
Authority
CN
China
Prior art keywords
signals
load
power
utility system
inference
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202080041162.3A
Other languages
Chinese (zh)
Inventor
C·T·艾福德三世
K·C·格罗斯
王光超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Oracle International Corp
Original Assignee
Oracle International Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Oracle International Corp filed Critical Oracle International Corp
Publication of CN114175071A publication Critical patent/CN114175071A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/04—Circuit arrangements for ac mains or ac distribution networks for connecting networks of the same frequency but supplied from different sources
    • H02J3/06—Controlling transfer of power between connected networks; Controlling sharing of load between connected networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0206—Price or cost determination based on market factors
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Game Theory and Decision Science (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Educational Administration (AREA)
  • Operations Research (AREA)
  • Health & Medical Sciences (AREA)
  • Power Engineering (AREA)
  • Data Mining & Analysis (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • General Health & Medical Sciences (AREA)
  • Water Supply & Treatment (AREA)
  • Supply And Distribution Of Alternating Current (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The system receives a set of load signals from a profile that contains historical load information collected at different locations throughout the grid that distributes power to the utility system. Next, the system applies a first order difference function to the set of load signals to generate a set of difference signals. The system then performs a spike detection operation on the set of differential signals to identify pairs of positive and negative and positive spikes that identify gaps in the set of load signals associated with network outage periods. Next, the system modifies the set of load signals by filling each identified gap with a predicted load value determined based on performing a local load shape prediction operation on consecutive load values immediately preceding the identified gap. Finally, the system predicts the power demand of the utility system based on the set of modified load signals.

Description

Intelligent data pre-processing techniques for facilitating load shape prediction for utility systems
Technical Field
The disclosed embodiments relate generally to techniques for performing power demand forecasting to facilitate ongoing operation of a utility system. More particularly, the disclosed embodiments relate to a smart load data preprocessing technique that facilitates improved electrical load shape prediction for utility systems.
Background
Power utility systems typically provide very limited storage for power, meaning that supply and demand must be continuously matched. Regional blackouts can occur when demand suddenly exceeds supply. To avoid such blackouts, utility companies are constantly buying and selling power through regional and national grids. Accurate load shape predictions for predicting demand for two or more hours into the future are critical, as real-time spot market rates for buying and selling power on the grid may vary by a factor of 10 to 20. For this reason, most utility systems begin to analyze information from a multi-year data historian (historian) profile and long-term historical weather patterns using machine learning techniques to create short-term load shape predictions based on current and recently predicted weather conditions. These short term forecasts allow optimization of supply and demand decisions to minimize the cost of spot market purchases and maximize the hourly revenue for spot market sales.
Utility systems also focus on long-term load shape prediction to anticipate future weeks and months of demand, where such demand is less affected by hourly or daily weather fluctuations, but more affected by anticipated population growth (or decline) patterns, seasonal weather patterns, and residential/commercial demand growth patterns in the geographic area served by the utility. Power companies use such long-term demand forecasts to perform important operations, such as: managing a demand side; storage maintenance and scheduling; integration of renewable energy sources; coordinating the supply of cheaper electricity by alternative means such as energy exchange; and enters into bilateral electricity supply agreements with nearby utilities and cogeneration facilities.
The accuracy of short-term and long-term predictions is strongly influenced by the fidelity of the archived signals in the data historian archive. However, the fidelity of such archived signals is often adversely affected by network outage events, such as transformer failures or periodic maintenance. Archived data generated during such network outage events is inconsistent with the normal operation of the utility system and can result in very inaccurate short-term and long-term demand forecasts.
Accordingly, there is a need for a technique for mitigating the adverse effects of anomalous disturbances on archived data historian signals generated during such network outage events.
Disclosure of Invention
The disclosed embodiments relate to a system for predicting power demand of a utility system. During operation, the system first receives a set of load signals from a profile containing historical load information collected at different locations throughout the grid that distributes power to the utility system. Next, the system pre-processes the set of load signals. During the preprocessing operation, the system applies a first order difference function to the set of load signals to generate a set of difference signals. The system then performs a spike detection operation on the set of differential signals to identify pairs of positive and negative and positive spikes that identify gaps in the set of load signals associated with network outage periods. Next, the system modifies the set of load signals by filling each identified gap with a predicted load value determined based on performing a local load shape prediction operation on consecutive load values immediately preceding the identified gap. After the pre-processing operation is complete, the system predicts the power demand of the utility system based on the set of pre-processed load signals.
In some embodiments, the system uses the prediction of power demand to control the supply of power provided by the utility system.
In some embodiments, controlling the supply of power provided by the utility system comprises one or more of: controlling an amount of electricity generated by one or more power plants in a utility system; purchasing power for a utility system through a power grid; selling electricity produced by a utility system through a power grid; storing the power for future use by the utility system; and planning for utility systems to build new power plants or to add other power generation assets (e.g., wind turbines, gas turbines, solar farms, or geothermal assets).
In some embodiments, in predicting the power demand of a utility system, the system trains an inference model using a set of input signals including a set of load signals and other input signals, the model learning correlations between the set of input signals. Next, the system generates a set of inference signals using the inference model, wherein the inference model generates an inference signal for each input signal in the set of input signals. The system then uses a fourier-based decomposition and reconstruction technique that decomposes each signal in the set of inference signals into deterministic and stochastic components, and uses the deterministic and stochastic components to generate a set of composite signals that are statistically indistinguishable from the inference signals. Finally, the system projects the combined signal into the future to generate a prediction of the power demand for the utility system.
In some embodiments, the inference model is trained using Multivariate State Estimation Techniques (MSET).
In some embodiments, in generating a set of composite signals using fourier-based decomposition and reconstruction techniques, the system uses a telemetry parameter synthesis (TPSS) technique that creates a high fidelity synthesis equation for generating the composite signals.
In some embodiments, in generating a set of composite signals, the system first generates a set of non-normalized signals. The system then performs an ambient weather normalization operation on the set of non-normalized signals to generate the combined signal, wherein the ambient weather normalization operation adjusts the set of non-normalized signals using the historical, current and predicted weather measurements, and the historical power usage data to account for the effect of weather on the prediction of power demand.
In some embodiments, the other input signals include electricity usage data from a group of smart meters, wherein each smart meter in the group collects electricity usage data from residential and commercial customers of the utility system.
In some embodiments, in replacing load values in identified gaps with predicted load values, the system uses an optimal interpolation technique that replaces missing load values in the set of load signals with interpolated load values determined based on correlations between the load signals.
Drawings
Fig. 1A illustrates an exemplary grid circuit including a left side circuit and a right side circuit in accordance with the disclosed embodiments.
Fig. 1B illustrates the same grid circuit with a transformer moved from the right-side circuit to the left-side circuit, in accordance with the disclosed embodiments.
Fig. 1C illustrates the same grid circuit with an isolated fault near the transformer, in accordance with the disclosed embodiments.
FIG. 2A presents a graph illustrating a load pattern for a load shifting scenario in accordance with the disclosed embodiments.
FIG. 2B presents a graph illustrating a load pattern for a power outage situation in accordance with the disclosed embodiments.
FIG. 3 illustrates a power utility system including a group of power generation stations connected to homes and businesses through a power grid in accordance with the disclosed embodiments.
FIG. 4 presents a flowchart illustrating how a load shape prediction is computed in accordance with the disclosed embodiments.
FIG. 5 presents a flowchart of a technique of pre-processing load data and then using the pre-processed load data to predict power demand in accordance with a disclosed embodiment.
FIG. 6 presents a diagram illustrating a preprocessor in accordance with the disclosed embodiments.
FIG. 7 presents a flowchart illustrating a process for forecasting power demand in accordance with the disclosed embodiments.
Fig. 8 presents a flowchart illustrating a process of replacing a load value in a gap associated with a network outage period with a predicted load value in accordance with a disclosed embodiment.
Detailed Description
The following description is presented to enable any person skilled in the art to make and use the embodiments, and is provided in the context of a particular application and its requirements. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the present embodiments. Thus, the present embodiments are not limited to the embodiments shown, but are to be accorded the widest scope consistent with the principles and features disclosed herein.
The data structures and code described in this detailed description are typically stored on a computer-readable storage medium, which may be any device or medium that can store code and/or data for use by a computer system. Computer-readable storage media include, but are not limited to, volatile memory, non-volatile memory, magnetic and optical storage devices such as disk drives, magnetic tape, CDs (compact discs), DVDs (digital versatile discs or digital video discs), or other media capable of storing computer-readable media now known or later developed.
The methods and processes described in the detailed description section can be embodied as code and/or data, which can be stored in a computer-readable storage medium as described above. When a computer system reads and executes the code and/or data stored on the computer-readable storage medium, the computer system performs the methods and processes embodied as data structures and code and stored in the computer-readable storage medium. Furthermore, the methods and processes described below may be embodied in hardware modules. For example, hardware modules may include, but are not limited to, Application Specific Integrated Circuit (ASIC) chips, Field Programmable Gate Arrays (FPGAs), and other programmable logic devices now known or later developed. When the hardware modules are activated, the hardware modules perform the methods and processes contained in the hardware modules.
SUMMARY
To understand the way in which an electrical load can be interrupted, we first examine two common scenarios. Consider fig. 1A, where two circuits originating from the left and right sides of the figure share a common tie switch 102. These two circuits typically carry a 200kVA load. Now assume that the closed midpoint switch 104 in the right circuit requires maintenance, so the transformer 106 is transferred from the right circuit to the left circuit by closing tie switch 102 and opening tie switch 104, as shown in fig. 1B. At this point, the left side circuit carries 300kVA, while the right side circuit only carries 100 kVA. In addition, all operations are normal without power failure. However, the two circular switches 102 and 104 are in an abnormal state (not an "invalid state", but an "abnormal configuration"), which results in the load levels of the two circuits being invalid for load prediction purposes. The left circuit does not see a permanent 50% load increase, nor should we project this load increase into the future. Similarly, the left circuit does not see its half-load permanently extinguished; it is moved temporarily to another circuit and will move back, possibly within a day or two. The corresponding load patterns for the load transfer cases of the left and right side circuits are illustrated in fig. 2A. Because we really want to use only "steady state" loading to determine growth rate and other changes, we want to be able to screen out these anomalies autonomously.
A similar thing happens during the power down condition shown in fig. 1C. In this case, there is an isolation fault near the transformer 106 that no longer receives power. The left circuit operates normally, but the right circuit only provides a 100kVA load when the outage is restored. Fig. 2B illustrates the corresponding load patterns for this power down condition for the left and right side circuits. Note that for a power outage situation, there may be some noise in the load signal depending on what the data historian picks up. For example, there may be a large current spike indicating the presence of a short circuit that would be interrupted by the circuit breaker, causing the current to temporarily drop to zero. However, the spike duration may not be sufficient for a utility's analog system control and data acquisition (SCADA) system to pick it up. Thus, after interrupting the fault to allow the arc to extinguish, the circuit will typically remain off-line for a few seconds, meaning that the sensor will see at least several zero current intervals. Once the fault is localized and the first sectionalizer is open, the first half of the circuit can be quickly restored, with the remainder coming back on-line after the fault is repaired. Also, from a predictive point of view, the circuit does not experience a "true" drop in load; it is simply a temporary loss of load value due to a power outage.
For each of the above scenarios, we want to identify where these anomalies occur in the data historian database so that we can "analyze the detuning (analysis de-tune)" anomaly patterns from the database before initiating the training process for short-term and long-term load shape prediction. During this prediction process, we want to calculate potential long-term trends, but these typically result in variations of only a few percent or less over the year, while short-term anomalies can be orders of magnitude larger and can easily mask long-term trends.
While the above example provides a simple illustration of two data cleansing scenarios using a simple line graph, in practice, the actual data historian time series graph is very dynamic with 30% diurnal (night to day) load variation in most large cities. These diurnal load variations are then superimposed on the extra 80% of the long-term seasonal variations in the hottest and coldest regions of the country (coldest winter to hottest summer). Thus, an automatic anomaly discovery process for facilitating pre-processing of training data for load shape prediction does not work efficiently simply by setting a threshold and then inferring network anomalies when the load exceeds the threshold. There is a need for an advanced statistical pattern recognition based technique to efficiently detect what we call "box singularities" (i.e., upward or downward square deviations superimposed on complex dynamic load shape patterns).
Our new technique for intelligent autonomous preprocessing screens long-term data historian signals and identifies these bin singularities in other steady-state time series signals. Our new technique first computes a first order difference function for each signal in the data historian, where the first order difference function is a numerical approximation of the first derivative of the time series signal. Note that the first order difference function highlights all the plateau (plateau) regions regardless of their magnitude. We can then identify and characterize the "square wave" deviation by setting a simple criterion for the peak in the first order difference function. For example, an upward rectangle in the load signal will include a positive spike and a subsequent negative spike in the first difference function, and a downward rectangle in the load signal will include a negative spike and a subsequent positive spike in the first difference function.
Note that we use a "spike detection" technique to detect such positive and negative spikes. (see, for example, the spike detection technique described in U.S. patent application No.16/215,345 entitled "synthetic High-Fidelity Signals with Spikes for promagnetic-Surveillance Applications" filed by the inventors of Guang C.Wang and Kenny C.Gross on 12/10.2018, which is hereby incorporated by reference). This spike detection technique is applied to the first difference function to detect pairs of positive and negative and positive spikes, which allows for the filtering of bin singularities from the data historian signals to facilitate short-term and long-term load shape prediction.
After all bin singularities are identified in the data historian signal, rather than merely cutting and discarding those bin singularities, the gaps resulting from the bin singularities are filled with load values inferred from previous load values. For each bin singularity, we first (1) extract the longest possible piece of "normal" data that occurred before the bin singularity. (2) We then use the extracted segments to train a "miniature predictive model" for the load shape prediction technique. (see, for example, the load shape prediction model described in U.S. patent application No.15/715,692 entitled "Electric load shape forming Based on smart meters Signals" filed by inventors Kenny c.gross, singing Li, and Guan c.wang at 2017, 26.9.9.78, which is hereby incorporated by reference). (3) Finally, we use this miniature predictive model to predict load values over the box singularity time span.
For example, if the network experiences no singularity for 18 months and then an interruption that requires two weeks to repair, then there will be two weeks of box singularities in the data stream in the data historian archive. In this case we used the load value 18 months in succession before the network outage to predict the load value during the two week outage period. This is advantageous over simply discarding data in the singularity of the bin to generate training data for load shape prediction. After all gaps in the load data associated with the box singularity are filled with the inferred load values, we train the load shape prediction model to project the electrical load into the future.
Before further describing our pretreatment techniques, we first describe the utility system in which it operates.
Exemplary Utility System
FIG. 3 illustrates an exemplary utility system including a group of power plants 302 and 304 connected to homes and businesses 310 through a power grid 306, in accordance with the disclosed embodiments. Note that power plant 302 and 304 may generally include any type of power generation facility, such as a nuclear power plant, a solar power plant, a windmill or windpark, or a coal, gas, or oil fired power plant. The power plant 302 and 304 are connected to a power grid 306, and the power grid 306 transmits power to homes and businesses 310 within the area served by the utility system 300, and also transmits power to and from other utility systems. Note that the power grid 306 transmits power to homes and businesses 310 through individual smart meters 308, and the smart meters 308 periodically transmit AMI signals containing power usage data (including kilowatt measurements and kilowatt-hour measurements) to the data center 320.
The control system within the data center 320 receives the AMI signal from the smart meter 308 and the weather data 312, including historical, current, and predicted weather information, and generates a load forecast that is used to send a control signal 325 to the power plant 302 and the power grid 306. During operation of the system, load data 327 from the power grid 306 is received by the data center 320 and stored in the data historian profile 330. This load data is then used to optimize load prediction, as described in more detail below.
Generating load shape predictions
FIG. 4 presents a flowchart illustrating how the above-described system calculates the optimal load shape prediction 418 in accordance with the disclosed embodiments. The system begins with an AMI meter signal 402 obtained from a number of smart meters in a utility system. As shown in fig. 4, these AMI meter signals 402 include a historical AMI signal 403 and a recent AMI signal 404. The system feeds the recent AMI signal 404 to an inference MSET module 405 that trains an inference model to learn correlations between the recent AMI signal 404, and then generates a set of inference signals 406 using the trained inference model. Next, the system feeds the inference signals 406 to a TPSS synthesis module 408 that performs a TPSS training operation 410 that decomposes each signal in the set of inference signals 406 into deterministic and stochastic components, and then uses the deterministic and stochastic components to generate a corresponding set of synthetic signals that are statistically indistinguishable from the inference signals. Finally, the system projects the combined signal into the future to generate an unnormalized TPSS forecast 412 for the power demand of a group of utility customers.
Next, the system feeds the non-normalized TPSS predictions 412 to an ambient weather normalization module 416, which normalizes the non-normalized TPSS predictions 412 to account for power usage changes caused by predicted ambient weather changes. The normalization process involves analyzing the historical AMI signal 403 against the historical weather measurements 414 to determine how the AMI meter signal 402 varies for different weather patterns. The normalization process then uses the current and predicted weather measurements 415 to modify the un-normalized TPSS predictions 412 to account for the predicted weather conditions. This results in a final load shape prediction 418 that the utility system can use to perform various operations as described above to control the power supply provided by the utility system.
Predicting power demand
FIG. 5 presents a flowchart of a technique of pre-processing load data and then using the pre-processed load data to predict power demand in accordance with a disclosed embodiment. The system first receives a set of load signals from a profile that contains historical load information collected at various locations throughout the grid that distributes power to a utility system (step 502). Next, the system applies a first difference function to the set of load signals to generate a set of difference signals (step 504). The system then performs a spike detection operation on the set of difference signals to identify pairs of positive and negative and positive spikes that identify gaps in the set of load signals associated with network outage periods (step 506). Next, the system modifies the set of load signals by filling each identified gap with a predicted load value determined based on performing a local load shape prediction operation on consecutive load values immediately preceding the identified gap (step 508). The system then predicts the power demand of the utility system based on the set of modified load signals (step 510). Finally, the system uses the prediction of the demand for electricity to control the supply of electricity provided by the utility system (step 512). Note that steps 504, 506, and 508 are performed by a preprocessor 604 as shown in fig. 6, the preprocessor 604 preprocessing the data historian load signal 602 to produce a preprocessed data historian load signal 606.
FIG. 7 presents a flowchart illustrating a process for forecasting power demand in accordance with the disclosed embodiments. (the flowchart illustrates in more detail the operations performed in step 510 of the flowchart in figure 5). First, the system trains an inference model using a set of input signals, including a set of load signals and other input signals, that learns the correlations between the set of input signals (step 702). Next, the system generates a set of inference signals using the inference model, where the inference model generates an inference signal for each input signal in the set of input signals (step 704). The system then generates a set of non-normalized composite signals that are statistically indistinguishable from the inference signals using Fourier-based decomposition and reconstruction techniques that decompose each signal in the set of inference signals into deterministic and stochastic components and use the deterministic and stochastic components (step 706). Next, the system performs an ambient weather normalization operation on the set of un-normalized composite signals to generate the composite signal, wherein the ambient weather normalization operation adjusts the set of un-normalized signals to account for the predicted impact of weather on power demand using the historical, current and predicted weather measurements, and historical power usage data (step 708). Finally, the system projects the combined signal into the future to generate a prediction of the utility system power demand (step 710).
Fig. 8 presents a flowchart illustrating a process of replacing a load value associated with a network outage period in a gap with a predicted load value in accordance with a disclosed embodiment. (the flowchart illustrates in more detail the operations performed in step 508 of the flowchart in figure 5). First, the system initializes a loop counter i to 1 (step 802). Next, the system determines a cycle counter i<N (step 804). If not (NO at step 804), then the process is complete. If so (YES at step 804), then the system collects gapi-1And gapiWith successive load values in between (step 806). Next, the system builds a miniature load shape prediction model based on the collected load values to generate a gapiLoad value (step 808). The system then generates a gap using the miniature load shape prediction modeliThe load value of (step 810). Finally, the system populates the gap with the generated load valuei(step 812) and increments the loop counter i +1 (step 814) before returning to step 804.
Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.
The foregoing description of the embodiments has been presented for the purposes of illustration and description only. They are not intended to be exhaustive or to limit the specification to the forms disclosed. Accordingly, many modifications and variations will be apparent to practitioners skilled in the art. Furthermore, the above disclosure is not intended to limit the present description. The scope of the present description is defined by the appended claims.

Claims (20)

1. A method for predicting power demand of a utility system, comprising:
receiving a set of load signals from a profile containing historical load information collected at various locations throughout a power grid distributing power for a utility system;
the set of load signals is pre-processed in the following way,
applying a first order difference function to the set of load signals to generate a set of differential signals,
performing spike detection operations on the set of differential signals to identify pairs of positive and negative positive spikes that identify gaps in the set of load signals associated with network outage periods, an
Modifying the set of load signals by filling each identified gap with a predicted load value determined based on performing a local load shape prediction operation on consecutive load values immediately preceding the identified gap; and
predicting a power demand of the utility system based on the set of preprocessed load signals.
2. The method of claim 1, wherein the method further comprises using the prediction of the demand for electricity to control the supply of electricity provided by the utility system.
3. The method of claim 2, wherein controlling the supply of power provided by the utility system comprises one or more of:
controlling an amount of electricity generated by one or more power plants in a utility system;
purchasing power for a utility system through a power grid;
selling power generated by a utility system through a power grid;
storing the power for future use by the utility system; and
plans are made for utility systems to build new power plants.
4. The method of claim 1, wherein predicting the power demand of the utility system comprises:
training an inference model using a set of input signals including the set of load signals and other input signals, the inference model learning correlations between the set of input signals;
generating a set of inference signals using the inference model, wherein the inference model generates inference signals for each input signal in the set of input signals;
generating a set of composite signals that are statistically indistinguishable from the inference signals using a Fourier-based decomposition and reconstruction technique that decomposes each signal in the set of inference signals into deterministic and stochastic components and uses the deterministic and stochastic components; and
projecting the set of composite signals into the future to generate a prediction of power demand for the utility system.
5. The method of claim 4, wherein the inference model is trained using Multivariate State Estimation Techniques (MSET).
6. The method of claim 4, wherein generating the set of synthetic signals using a fourier-based decomposition and reconstruction technique comprises using a telemetry parameter synthesis (TPSS) technique that creates a high fidelity synthesis equation for generating the set of synthetic signals.
7. The method of claim 4, wherein generating the set of composite signals comprises:
generating a set of non-normalized signals; and
performing an ambient weather normalization operation on the set of non-normalized signals to generate the set of composite signals, wherein the ambient weather normalization operation adjusts the set of non-normalized signals using historical, current and predicted weather measurements, and historical power usage data to account for weather effects on the power demand forecast.
8. The method of claim 4, wherein the other input signals include electricity usage data from a group of smart meters, wherein each smart meter in the group of smart meters collects electricity usage data from customers of a utility system.
9. The method of claim 1, wherein replacing the load values in the identified gaps with predicted load values involves using an optimal interpolation technique that replaces missing load values in the set of load signals with interpolated load values determined based on the correlation between the load signals.
10. A non-transitory computer-readable storage medium storing instructions that, when executed by a computer, cause the computer to perform a method for predicting power demand of a utility system, the method comprising:
receiving a set of load signals from a profile containing historical load information collected at various locations throughout a power grid distributing power for a utility system;
the set of load signals is pre-processed in the following way,
applying a first order difference function to the set of load signals to generate a set of differential signals,
performing spike detection operations on the set of differential signals to identify pairs of positive and negative positive spikes that identify gaps in the set of load signals associated with network outage periods, an
Modifying the set of load signals by filling each identified gap with a predicted load value determined based on performing a local load shape prediction operation on consecutive load values immediately preceding the identified gap; and
predicting a power demand of the utility system based on the set of preprocessed load signals.
11. The non-transitory computer readable storage medium of claim 10, wherein the method further comprises using the prediction of the demand for electricity to control the supply of electricity provided by the utility system.
12. The non-transitory computer-readable storage medium of claim 11, wherein controlling the supply of power provided by a utility system comprises one or more of:
controlling an amount of electricity generated by one or more power plants in a utility system;
purchasing power for a utility system through a power grid;
selling power generated by a utility system through a power grid;
storing the power for future use by the utility system; and
plans are made for utility systems to build new power plants.
13. The non-transitory computer readable storage medium of claim 10, wherein predicting the power demand of the utility system comprises:
training an inference model using a set of input signals including the set of load signals and other input signals, the inference model learning correlations between the set of input signals;
generating a set of inference signals using the inference model, wherein the inference model generates inference signals for each input signal in the set of input signals;
generating a set of composite signals that are statistically indistinguishable from the inference signals using a Fourier-based decomposition and reconstruction technique that decomposes each signal in the set of inference signals into deterministic and stochastic components and uses the deterministic and stochastic components; and
projecting the set of composite signals into the future to generate a prediction of power demand for the utility system.
14. The non-transitory computer-readable storage medium of claim 13, wherein the inference model is trained using Multivariate State Estimation Techniques (MSET).
15. The non-transitory computer readable storage medium of claim 13, wherein generating the set of synthetic signals using a fourier-based decomposition and reconstruction technique comprises using a telemetry parameter synthesis (TPSS) technique that creates a high fidelity synthesis equation for generating the set of synthetic signals.
16. The non-transitory computer-readable storage medium of claim 13, wherein generating the set of composite signals comprises:
generating a set of non-normalized signals; and
performing an ambient weather normalization operation on the set of non-normalized signals to generate the set of composite signals, wherein the ambient weather normalization operation adjusts the set of non-normalized signals using historical, current and predicted weather measurements, and historical power usage data to account for weather effects on the power demand forecast.
17. The non-transitory computer readable storage medium of claim 13, wherein the other input signals comprise electricity usage data from a set of smart meters, wherein each smart meter in the set of smart meters collects electricity usage data from customers of a utility system.
18. The non-transitory computer readable storage medium of claim 10, wherein replacing the load values in the identified gaps with predicted load values involves using an optimal value interpolation technique that replaces missing load values in the set of load signals with interpolated load values determined based on correlations between the load signals.
19. A system for predicting power demand of a utility system, comprising:
at least one processor and at least one associated memory; and
a prediction mechanism executing on the at least one processor, wherein during operation the prediction mechanism:
receiving a set of load signals from a profile containing historical load information collected at various locations throughout a power grid distributing power for a utility system;
applying a first order difference function to the set of load signals to generate a set of differential signals,
performing spike detection operations on the set of differential signals to identify pairs of positive and negative and positive and negative spikes that identify gaps in the set of load signals associated with network outage periods,
modifying the set of load signals by filling each identified gap with a predicted load value determined based on performing a local load shape prediction operation on consecutive load values immediately preceding the identified gap, an
Predicting a power demand of the utility system based on the modified set of load signals.
20. The system of claim 19, wherein the system additionally uses the prediction of power demand to control the supply of power provided by a utility system.
CN202080041162.3A 2019-07-01 2020-05-05 Intelligent data pre-processing techniques for facilitating load shape prediction for utility systems Pending CN114175071A (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US16/458,498 2019-07-01
US16/458,498 US10862302B1 (en) 2019-07-01 2019-07-01 Intelligent data preprocessing technique to facilitate loadshape forecasting for a utility system
PCT/US2020/031474 WO2021002930A1 (en) 2019-07-01 2020-05-05 Intelligent data preprocessing technique to facilitate loadshape forecasting for a utility system

Publications (1)

Publication Number Publication Date
CN114175071A true CN114175071A (en) 2022-03-11

Family

ID=70740823

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202080041162.3A Pending CN114175071A (en) 2019-07-01 2020-05-05 Intelligent data pre-processing techniques for facilitating load shape prediction for utility systems

Country Status (5)

Country Link
US (1) US10862302B1 (en)
EP (1) EP3994637A1 (en)
JP (1) JP2022539123A (en)
CN (1) CN114175071A (en)
WO (1) WO2021002930A1 (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11941676B2 (en) * 2019-12-17 2024-03-26 Ebay Inc. Automatic ordering of consumable items
CA3206981A1 (en) * 2020-12-31 2022-07-07 Schneider Electric Systems Usa, Inc. Systems and methods for benchmarking operator performance for an industrial operation
CN116404745B (en) * 2023-02-27 2023-10-03 合肥威标科技有限公司 Intelligent regulation and control auxiliary decision-making system and method for power distribution network based on big data
CN116431355B (en) * 2023-06-13 2023-08-22 方心科技股份有限公司 Computing load prediction method and system based on power field super computing platform
US12009660B1 (en) 2023-07-11 2024-06-11 T-Mobile Usa, Inc. Predicting space, power, and cooling capacity of a facility to optimize energy usage
US11899516B1 (en) 2023-07-13 2024-02-13 T-Mobile Usa, Inc. Creation of a digital twin for auto-discovery of hierarchy in power monitoring

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040225625A1 (en) 2003-02-07 2004-11-11 Van Gorp John Christopher Method and system for calculating and distributing utility costs
US7058522B2 (en) 2003-05-13 2006-06-06 Siemens Power Transmission & Distribution, Inc. Very short term load prediction
WO2012138688A1 (en) * 2011-04-04 2012-10-11 The Catholic University Of America Systems and methods for improving the accuracy of day-ahead load forecasts on an electric utility grid
US10043224B2 (en) * 2012-08-10 2018-08-07 Itron, Inc. Unified framework for electrical load forecasting
US20150120224A1 (en) * 2013-10-29 2015-04-30 C3 Energy, Inc. Systems and methods for processing data relating to energy usage
US20150161233A1 (en) 2013-12-11 2015-06-11 The Board Of Trustees Of The Leland Stanford Junior University Customer energy consumption segmentation using time-series data
JP7051856B2 (en) * 2016-11-29 2022-04-11 ピーク パワー インコーポレイテッド Systems and methods for controlling dynamic energy storage systems
US10310459B2 (en) 2017-09-26 2019-06-04 Oracle International Corporation Electric loadshape forecasting based on smart meter signals
US10734811B2 (en) * 2017-11-27 2020-08-04 Ihi Inc. System and method for optimal control of energy storage system

Also Published As

Publication number Publication date
EP3994637A1 (en) 2022-05-11
WO2021002930A1 (en) 2021-01-07
US10862302B1 (en) 2020-12-08
JP2022539123A (en) 2022-09-07

Similar Documents

Publication Publication Date Title
CN114175071A (en) Intelligent data pre-processing techniques for facilitating load shape prediction for utility systems
JP7180014B2 (en) Power system control using dynamic power flow model
US10443577B2 (en) Systems and methods for improved wind power generation
Kuznetsova et al. Analysis of robust optimization for decentralized microgrid energy management under uncertainty
US9002530B2 (en) Power plant operation enhancement
Hamidieh et al. Microgrids and resilience: A review
US11575263B2 (en) Optimal power flow control via dynamic power flow modeling
Ferraro et al. From Uninterruptible Power Supply to resilient smart micro grid: The case of a battery storage at telecommunication station
JP2017517042A (en) System and method for utility worker prediction
WO2020171814A1 (en) Event selection for power grid disturbance
US11451053B2 (en) Method and arrangement for estimating a grid state of a power distribution grid
JP2014219970A (en) Systems and methods for estimating reliability return on utility vegetation management
Hartmann et al. Island mode operation in intelligent microgrid—Extensive analysis of a case study
US20220284156A1 (en) Real-time update of power system models for dynamic security assessment
Reponen et al. Distributed automation solution and voltage control in MV and LV distribution networks
Lu et al. Smart power systems and smart grids: toward multi-objective optimization in dispatching
Onen Model-based grid modernization economic evaluation framework
Poudel Quantifying and improving resilience in power distribution systems
Losi et al. Distribution control center: New requirements and functionalities
Buechner et al. Automated network reinforcement including a model for an asset management strategy
JPWO2021002930A5 (en)
Aıroboman Reliability Assessment of Power System Network: A Detailed Review
Chang Risk Assessment and Optimal Response Strategies for Resilience of Electric Power Infrastructure to Extreme Weather
Schoene et al. Operational Forecasting–Use Cases and Implementation Challenges
Adedokun et al. Machine Learning-Based Restoration Forecast with Predictive Power Outage for Diverse Power Outage Scenarios

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination